Omicron, our compact ultra-low-distortion headphone amplifier, includes a simple, compact, accurate, flexible, effective, reliable and affordable headphone protection against DC voltages and turn on/off transients. Since there has been interest in using Omicron's protection with other headamps, we implemented the Omicron protection as a separate board:
The board is 48×48mm (1⅞ inch), mounting holes are in the corners of a 40×40mm square, all parts are easily available and through hole, all connections are routed in a single copper layer.
Details will follow.
The board is 48×48mm (1⅞ inch), mounting holes are in the corners of a 40×40mm square, all parts are easily available and through hole, all connections are routed in a single copper layer.
Details will follow.
Last edited:
The schematic is virtually identical to the one published last year, with a couple of resistors and a ground connect jumper added:
The signal from each channel is low-pass filtered with R9 C5 R11 C3 and R8 C6 R10 C4 and fed to a window comparator. If the DC component of the input signal falls outside of the window set by R2 R3 R4 R5 (+/-80-mV with the values shown and +/-17V power supply), the comparator switches off the Schmitt trigger T2T3, which turns off the relay and disconnects the load. The protection is fast - a 1V step at the input disconnects the load in 30 milliseconds.
When DC disappears, C7 slowly charges via R12, so the relay turns on with a delay of about 1 second with the values shown. The same delay protects the load from turn-on transients.
T1 optionally provides fast turn-off, in cooperation with the power supply. The gate of T1 is connected to a rectifier with a small filter capacitor along these lines (Omicron power supply includes a suitable rectifier):
Normally, the gate is at or below the negative rail (-17V) but is pulled to ground by Rbleed when AC power disappears, which discharges C7 and turnes off the relay.
The part list is simple, with just 12 lines and 30 components in total (excluding the board and connectors). Most likely, you have most if not all parts already. The total cost at today's prices on Mouser is less than $8, with the relay being the most expensive component at $3.
The signal from each channel is low-pass filtered with R9 C5 R11 C3 and R8 C6 R10 C4 and fed to a window comparator. If the DC component of the input signal falls outside of the window set by R2 R3 R4 R5 (+/-80-mV with the values shown and +/-17V power supply), the comparator switches off the Schmitt trigger T2T3, which turns off the relay and disconnects the load. The protection is fast - a 1V step at the input disconnects the load in 30 milliseconds.
When DC disappears, C7 slowly charges via R12, so the relay turns on with a delay of about 1 second with the values shown. The same delay protects the load from turn-on transients.
T1 optionally provides fast turn-off, in cooperation with the power supply. The gate of T1 is connected to a rectifier with a small filter capacitor along these lines (Omicron power supply includes a suitable rectifier):
Normally, the gate is at or below the negative rail (-17V) but is pulled to ground by Rbleed when AC power disappears, which discharges C7 and turnes off the relay.
The part list is simple, with just 12 lines and 30 components in total (excluding the board and connectors). Most likely, you have most if not all parts already. The total cost at today's prices on Mouser is less than $8, with the relay being the most expensive component at $3.
For BTL aka "balanced" amplifiers, you'd need a different DC detector, sensitive to differential voltage. There is an excellent example elsewhere on this forum. Out of curiosity, what balanced headamp did you build?
If you use a DC blocking capacitor, you need no DC protection, although may still need protection against turn-on thump. However, a protection circuit is totally transparent until a fault happens, while a big coupling capacitor is not. Such capacitor would be out of place in any amplifier with split supply rails.
If you use a DC blocking capacitor, you need no DC protection, although may still need protection against turn-on thump. However, a protection circuit is totally transparent until a fault happens, while a big coupling capacitor is not. Such capacitor would be out of place in any amplifier with split supply rails.
I thought well designed headphone amplifier does not need protection. Headphones are not demanding any big power, few milliwatts is all it takes, and even if the design is for a watt, thats negligable power, not much current, most reasonable headphones are 50 ohms and higher, nothing like classA amps. All it takes is one cap on the output. I have many great sounding cap coupled headphone amps, why such aversion against caps? There are literally thousands in recording studio.
With a split rails headamp, one can take any of the three paths: (1) keep fingers crossed and hope no fault can ever happen (off topic: my oven stopped working after a thunderstorm today); (2) add a big cap at the output just in case something does happen; (3) add a relay and a few other parts to protect your cans in case of a fault. I am not averse to any of these choices; it's your cans on the line, not mine. Should you choose (3), the schematic and board above may be helpful.
Last edited:
hello.
what is the function
1. "GND Connect" pin? why not directly connect to ground? why there is "option"
2. R6 and R7?
3. if am using 12v coil relay can i adjust R16? or something else to be adjusted? and at what value for 12v coil relay?
thank you so much
1. The option to connect the power ground to the signal ground adds some flexibility when integrating the protection into an amplifier. Normally, there should be no connection, as the power ground should be connected to the power supply's ground, while the signal ground - to the amplifier's output ground.
2. R6 and R7 sink the comparators' input bias currents. These resistors are not required if the protection circuit is permanently connected to an amplifier's output (as in the Omicron). However, should the amplifier's output be unconnected and R6/R7 not installed, the input bias currents would slowly charge C3 C4 C5 C6, triggering protection.
3. R16 drops the difference between the rail-to-rail power supply voltage and the relay's coil voltage rating. The value, obviously, depends on the impedance of your relay's coil.
For example, for Fujitsu RY-12W-K the coil resistance is 960 ohm, which corresponds to 12V/960ohm=12.5mA. If your rails are +/-15V, you'd need to drop (15+15)-12=18V, and R16 would be 18V/12.5mA=1.44kOhm. From the E24 list of standard values, 1.5kOhm would work well. Note that R16 would dissipate 18V×12.5mA=225mW, so you'd need at least a 0.5W resistor, and 1W would be even better. For a lower impedance coil, the current and the power dissipation would be proportionally higher, and the required resistance would be lower.
2. R6 and R7 sink the comparators' input bias currents. These resistors are not required if the protection circuit is permanently connected to an amplifier's output (as in the Omicron). However, should the amplifier's output be unconnected and R6/R7 not installed, the input bias currents would slowly charge C3 C4 C5 C6, triggering protection.
3. R16 drops the difference between the rail-to-rail power supply voltage and the relay's coil voltage rating. The value, obviously, depends on the impedance of your relay's coil.
For example, for Fujitsu RY-12W-K the coil resistance is 960 ohm, which corresponds to 12V/960ohm=12.5mA. If your rails are +/-15V, you'd need to drop (15+15)-12=18V, and R16 would be 18V/12.5mA=1.44kOhm. From the E24 list of standard values, 1.5kOhm would work well. Note that R16 would dissipate 18V×12.5mA=225mW, so you'd need at least a 0.5W resistor, and 1W would be even better. For a lower impedance coil, the current and the power dissipation would be proportionally higher, and the required resistance would be lower.
The values shown work for 18VAC secondaries. You can make it work with pretty much any voltage, provided that the gate of the MOSFET never goes above or below its source (=negative supply rail) by more than the MOSFET gate-to-source voltage rating (+/-20V typical). To achieve that, choose R1 R13 to make the MOSFET safe, then choose Rbleed so that MOSFET stays off when AC is present but opens quickly when AC disappears.
Not for sale yet, but will be soon.
Hi Alex, what a wonderful and creative circuit. Being already a fan of your omicron i think i found my next project.
My omicron iteration is dual mono, so this circuit will have to be split into single channels.
For that, which comparator should i use?
And is there any way to simplify or eliminate T1's separate psu requirement? My main concern for it is the layout challenge
Lastly how about flipping the relay connection so that when its off it connects the output to the ground and at ON its a lift, effectively removing it from the signal path? Its how myref amp's relay circuit is done and i like it.
Thanks
My omicron iteration is dual mono, so this circuit will have to be split into single channels.
For that, which comparator should i use?
And is there any way to simplify or eliminate T1's separate psu requirement? My main concern for it is the layout challenge
Lastly how about flipping the relay connection so that when its off it connects the output to the ground and at ON its a lift, effectively removing it from the signal path? Its how myref amp's relay circuit is done and i like it.
Thanks
For a single channel, you'd need a dual comparator with open collector/drain outputs and the maximum supply voltage above what you're going to supply it with. Something like the LM393 should work, and there are many other suitable parts.
T1's purpose is to quickly disconnect the load when AC power is lost. With the Omicron, you can simply omit T1 - Omicron shuts down quietly.
Unlike the LM3886, the Omicron's output stage in unprotected from overcurrent other than by the emitter resistors. It survives brief shorts such as when you plug in your headphones, but I would advise against shorting Omicron's output to the ground with a relay. BTW the specified relays do not add any distortion - all distortion measurement and listening tests for the Omicron were performed with the relay contacts in the circuit.
Not Mauro's MyRef: